integrating r and c/c++ - boston · pdf filercpp, rcpparmadillo, rcppbugs, etc. applications...

Integrating R and C/C++

Robert Putnam

Katia Oleinik

Information Services and Technology

Introduction

Background of participants, motivations

Agenda Day one

Linux/emacs basics

C hands-on tutorial

Day two

C wrap-up

C++ introduction

Calling C/C++ from R

Benchmarking, debugging, etc.

Rcpp, RcppArmadillo, RcppBUGS, etc.

Applications

Information Services & Technology 5/9/2012

Introduction to programming in C: Goals

To write simple C programs

To understand and modify existing C code

To write and use makefiles

Information Services & Technology

4

5/9/2012

C History

Developed by Dennis Ritchie at Bell Labs in 1969-73 Originally designed for system software

Impetus was porting of Unix to a DEC PDP-11

PDP-11 had 24kB main memory!

See The C Programming Language by Kernighan &

Ritchie (2nd ed.) (aka “K & R”)

Official ANSI standard published in 1989 Updated in 1999

C++ (1983) Author: Bjarne Stroustrup (Bell Labs), 1979, “C with classes”

More later…


5

5/9/2012

Compiled vs. Interpreted Languages

Interpreted languages when you type something, e.g., “x=y+z”, it is immediately parsed,

converted to machine language, and executed

examples: R, MATLAB, Python

advantage

interactive, allows fast development

disadvantage

generally uses more CPU/memory/time for given task


6

5/9/2012

Compiled (cont’d)

Compiled languages examples: C, C++, Fortran

source code must be processed through a compiler

checks for correct syntax and semantics

translates source code into assembly, then assembles (or calls

assembler) to produce machine code

passes machine code to linker, which creates executable

this is the file that you actually run

example: .exe file in Windows

default name in Unix/Linux: a.out


7

5/9/2012

Variables

Variables are declared to have a certain type.

Common types include: int

“integer”

number with no decimal places: -56, 857436

float, double

“floating-point”

number with decimal: 1.234, 4.0, 7.

float: single precision, 32 bits*, ~7 significant digits

double: double precision, 64 bits*, ~16 significant digits


8

5/9/2012

*on most computers

Variables (cont’d)

char

“character”

enclosed in single quotes

‘x’, ‘$’

character string is string of chars enclosed in double quotes

“This is a character string.”


9

5/9/2012

Syntax

C is case-sensitive

Spaces, linefeeds, etc., don’t matter except within

character strings.

Source lines end with semicolons (optional in R)

Comments notes for humans that are ignored by the compiler

C: enclosed by /* */

C++: // at beginning of comment

many C compilers also accept this syntax

Official advice: use them liberally (so you can still understand your

program next year [or next week, depending on your age])


10

5/9/2012

Functions

(As in R), source code largely consists of functions each one performs some task

you write some of them

some are supplied, typically in libraries

every code contains at least one function, called main

functions often, though not always, return a value, e.g.: int, float, char, etc.

default return value is int

To explicit about returning no value, declare as void


11

5/9/2012

Functions (cont’d)

functions may, but do not have to, take arguments “arguments” are inputs to the function

e.g., y = sin(x)

code blocks, including entire functions, are enclosed

within “curly brackets” { }

main function is defined in source code as follows:

int main( ) {

function statements

}


12

5/9/2012

type declaration function name function arguments

(we [currently]have no arguments here

but still need parentheses)

Functions (3)

Style note: some people like to arrange the brackets

like

int main( )

{

function statements

}

Either way is fine Friendly advice: be consistent!

Emacs advertisement: a good editor can do automatic

indentation, help you find matching brackets, etc.


13

5/9/2012

How to say “hello, world”: printf

printf is a function, part of C’s standard input/output

library, that is used to direct output to the screen, e.g.,

printf(“my string”);

The above syntax does not include a line feed. We

can add one with:

printf(“my string\n”); where \n is a special character representing LF


14

5/9/2012

printf and stdio.h

Some program elements, such as library functions like

printf, are declared in header files, aka “include files.”

Syntax:

#include <stdio.h> or

#include “stdio.h”

The contents of the named file are presented to the

compiler as if you had placed them directly in your

source file. In the case of printf, stdio.h informs the

compiler about the arguments it takes, so the compiler

can raise a warning or error if printf is called

incorrectly. More will be said about this later.


15

5/9/2012

Exercise 1

Write a “hello, world” program in an editor

Program should print a character string

General structure of code, in order: include the file “stdio.h”

define main function

use printf to print string to screen

Save it to the file name hello.c

solution


16

5/9/2012

http://scv.bu.edu/documentation/presentations/intro_to_c/ex1/

Compilation

A compiler is a program that reads source code and

converts it to a form usable by the computer/CPU, i.e.,

machine code

Code compiled for a given type of processor will not

generally run on other types AMD and Intel are compatible

We’ll use gcc, since it’s free and readily available


17

5/9/2012

Compilation (cont’d)

Compilers have numerous options Try ‘man gcc’, or see gcc compiler documentation at

http://gcc.gnu.org/onlinedocs/

gcc refers to the “GNU compiler collection,” which includes the C

compiler (gcc) and the C++ compiler (g++)

For now, we will simply use the –o option, which

allows you to specify the name of the resulting

executable


18

5/9/2012



Compilation (3)

In a Unix window:

gcc –o hello hello.c

“hello” is name of executable file (compiler output)

“hello.c” is source file name (compiler input)

Compile your code

If it simply returns a Unix prompt it worked

If you get error messages, read them carefully and see

if you can fix the source code and re-compile


19

5/9/2012

Compilation (4)

Once it compiles correctly, type the name of the

executable

./hello

at the Unix prompt, and it will run the program

should print the string to the screen


20

5/9/2012

Declarations

different variable types have different internal

representations, so CPUs use different machine

instructions for int, float, etc.

must tell compiler the type of every variable by

declaring them

example declarations:

int i, jmax, k_value;

float xval, elapsed_time;

char aletter, bletter;


21

5/9/2012

Arithmetic +, -, *, /

No power operator (see next bullet)

Math functions declared in math.h pow(x,y) raises x to the y power

sin, acos, tanh, exp, sqrt, etc.

for some compilers, need to add –lm flag (that’s a small el) to

compile command to link against math library

Exponential notation indicated by letter “e”

4.2e3

Good practice to use decimal points with floats, e.g.,

x = 1.0 rather than x = 1


22

5/9/2012

3102.4

Arithmetic (cont’d)

Computer math Value of variable on left of equals sign is replaced by value of

expression on right

Many legal statements are algebraically nonsensical, e.g.,

i = i + 1;


23

5/9/2012

Arithmetic (cont’d)

++ and -- operators these are equivalent:

i = i+1;

i++; always increments/decrements by 1

+= these are equivalent:

x = x + 46.3*y;

x += 46.3*y;


24

5/9/2012

Arithmetic (3)

Pure integer arithmetic truncates result!

5/2 = 2

2/5 = 0

Can convert types with cast operator

float xval;

int i, j;

xval = (float) i / (float) j;


25

5/9/2012

A Little More About printf

To print a value (as opposed to a literal string), must

specify a format

For now we will use %f for a float and %d for an int

Here’s an example of the syntax: printf(“My integer value is %d and my float value is %f \n”, ival, fval);

The values listed at the end of the printf statement will

be embedded at the locations of their respective

formats.


26

5/9/2012

Exercise 2

Write program to convert Celcius temperature to

Fahrenheit and print the result. Hard-wire the Celcius value to 100.0

We’ll make it an input value in a subsequent exercise

Don’t forget to declare all variables

Here’s the equation:

F = (9/5)C + 32

solution


27

5/9/2012


Address-of Operator

Every variable has an address in which it is stored in

memory

In C, we sometimes need to access the address of a

variable rather than its value Will go into more details when we discuss pointers

Address-of operator & returns address of specified

variable &ival returns the address of the variable ival

rarely need to know actual value of address, just need to use it


28

5/9/2012

scanf

reads from keyboard

2 arguments character string describing format

address of variable

must include stdio.h

example int ival;

scanf("%d", &ival);


29

5/9/2012

Exercise 3

Modify Celcius program to read value from keyboard Prompt for Celcius value using printf

Read value using scanf

Rest can remain the same as last exercise

solution


30

5/9/2012


Arrays

Declare arrays using [ ]

float x[100];

char a[25];

Array indices start at 0 (not 1, as in R) Declaration of x above creates locations for x[0] through x[99]

Multiple-dimensional arrays are declared as follows:

int a[10][20]; Note: In C, the last axis varies fastest; we say that C stores arrays in

“row-major” order. R, on the other hand, stores arrays in “column-

major” order.


31

5/9/2012

For Loop

for loop repeats calculation over range of indices

for(i=0; i<n; i++) {

a[i] = sqrt( pow(b[i],2) + pow(c[i],2) );

}

for statement has 3 parts: initialization

completion condition (i.e., if true, keep looping)

what to do after each iteration


34

5/9/2012

Exercise 4

Write program to: declare two float vectors of length 3

prompt for first vector and read values

prompt for second vector and read values

calculate dot product

print the result

Solution

Possible to use “redirection of standard input” to avoid

retyping each time: % echo 1 2 3 4 5 6 | dotprod


35

5/9/2012

i

i

i bac

3

1


Pointers

When you declare a variable, a location of appropriate

size is reserved in memory

When you do an assignment, the value is placed in

that memory location


37

5/9/2012

double x;

x = 3.2;

0

8

16

32

address

3.2

Pointers (cont.)

A pointer is a variable containing a memory address

Declared using *

double *p;

Often used in conjunction with address-of operator &

double x, *p;

p = &x;


38

5/9/2012

Pointers (3)

double x, *p;

p = &x;


39

5/9/2012

1040

1048

1056

1064

address

0

8

16

24

address

p 16 x

Pointers (4)

Depending on context, * can also be the dereferencing

operator Value stored in memory location pointed to by specified pointer

*p = 3.2;

Common error

double *p;

*p = 3.2;

double x, *p;

p = &x;

*p = 3.2;

Information Services & Technology 5/9/2012

Wrong! – p doesn’t have value yet

correct

Pointers (5)

The name of an array is actually a pointer to the

memory location of the first element a[100]

“a” is a pointer to the first element of the array (a[0])

These are equivalent: x[0] = 4.53;

*x = 4.53;


41

5/9/2012

Pointers (6)

If p is a pointer and n is an integer, the syntax p+n

means to advance the pointer by n locations*

These are therefore equivalent: x[4] = 4.53;

*(x+4) = 4.53;


42

5/9/2012

*i.e., for most machines, 4*n bytes for an int, and 8*n

bytes for a double

Pointers (7)


43

5/9/2012

In multi-dimensional arrays, values are stored in

memory with last index varying most rapidly

(a[0][0], a[0][1], a[0][2], … ) Opposite of R, MATLAB, et al.

The two statements in each box are equivalent for an

array declared as int a[5][5]:

a[0][3] = 7; a[1][0] = 7;

*(a+3) = 7; *(a+5) = 7;

sizeof


44

5/9/2012

Some functions require size of something in bytes

A useful function – sizeof(arg) The argument arg can be a variable, an array name, a type

Returns no. bytes in arg

double x, y[5];

sizeof(x) ( 8)

sizeof(y) (40)

sizeof(double) ( 8)

Dynamic Allocation

Suppose you need an array, but you don’t know how

big it needs to be until run time.

Tried and true method - use malloc function: malloc(n)

n is no. bytes to be allocated

returns pointer to allocated space

declared in stdlib.h

Many C compilers now accept “float f[n]”, where ‘n’ is

determined at runtime.


45

5/9/2012

Dynamic Allocation (cont’d)


46

5/9/2012

Declare pointer of required type float *myarray;

Suppose we need 101 elements in array: myarray = malloc(101*sizeof(float));

free releases space when it’s no longer needed: free(myarray);

Exercise 5

Modify dot-product program to handle vectors of any

length Prompt for length of vectors (printf)

Read length of vectors from screen (scanf)

Dynamically allocate vectors (malloc)

Prompt for and read vectors (printf, scanf)

use for loop

Don’t forget to include stdlib.h so you have access to the malloc

function declaration

solution


47

5/9/2012


if/else


48

5/9/2012

Conditional execution of block of source code

Based on relational operators < less than

> greater than

== equal

<= less than or equal

>= greater than or equal

!= not equal

&& and

|| or

if/else (cont’d)

Condition is enclosed in parentheses

Code block is enclosed in curly brackets

if( x > 0.0 && y > 0.0 ) {

printf(“x and y are both positive\n”);

z = x + y;

}


49

5/9/2012

if/else (3)


50

5/9/2012

Can have multiple conditions by using else if

if( x > 0.0 && y > 0.0 ) {

z = 1.0/(x+y);

} else if( x < 0.0 && y < 0.0 ) {

z = -1.0/(x+y);

} else {

printf(“Error condition\n”);

}

Exercise 6


51

5/9/2012

In dot product code, check if the magnitude of the dot

product is less than using the absolute value

function fabsf. If it is, print a warning message. With some compilers you need to include math.h for the fabsf

function. You should include it to be safe.

With some compilers you would need to link to the math library by

adding the flag –lm to the end of your compile/link command.

solution

610


Functions


52

5/9/2012

C functions return a single value

Return type should be declared (default is int)

Argument types must be declared

Sample function definition:

float sumsqr(float x, float y) {

float z;

z = x*x + y*y;

return z;

}

Functions (cont’d)


53

5/9/2012

Use of sumsqr function:

a = sumsqr(b,c);

Call by value when function is called, copies are made of the arguments

scope of copies is scope of function

after return from function, copies no longer exist

Functions (3)


54

5/9/2012

b = 2.0; c = 3.0;

a = sumsqr(b, c);

printf(“%f”, b);

float sumsqr(float x, float y) {

float z;

z = x*x + y*y;

x = 1938.6;

return z;

}

this line has no effect on b

will print 2.0

Functions (4)

If you want to change argument values, pass pointers

int swap(int *i, int *j) {

int k;

k = *i;

*i = *j;

*j = k;

return 0;

}


55

5/9/2012

Functions (5)

Let’s examine the following code fragment:

int a, b;

a = 2; b = 3;

swap(&a, &b);

Memory after setting values of a and b


56

5/9/2012

address

16

20

24

28

variable

b

a

3

2

Functions (6)

When function is called, copies of arguments are

created in memory

i, j are pointers to ints with values &a and &b


57

5/9/2012

address

16

20

24

28

variable

b

a

3

2

address

48

52

56

60

variable

j

i

24

20 &a i

&b j

swap(&a, &b); int swap(int *i, int *j){ ... }


58

5/9/2012

Functions (7)

What happens to memory for each line in the function?

k

address

16

20

24

28

variable

b

a

3

2

address

48

52

56

60

variable

j

i

24

20 int k;

address

16

20

24

28

variable

b

a

3

2

address

48

52

56

60

variable

j

i

24

20

k = *i;

k 2


59

5/9/2012

Functions (8)

k

address

16

20

24

28

variable

b

a

3

3

address

48

52

56

60

variable

j

i

24

20 *i = *j;

address

16

20

24

28

variable

b

a

2

3

address

48

52

56

60

variable

j

i

24

20

*j = k;

k 2

2


60

5/9/2012

Functions (9)

return 0;

address

16

20

24

28

variable

b

a

2

3

address

48

52

56

60

variable

24

20

2

Exercise 7

Modify dot-product program to use a function to

compute the dot product The function definition should go after the includes but before the

main program in the source file

Arguments can be an integer containing the length of the vectors and

a pointer to each vector

Function should only do dot product, no i/o

Do not give function same name as executable

I called my executable “dotprod” and the function “dp”

solution


61

5/9/2012


Function Prototypes

C compiler checks arguments in function definition and

calls number

type

If definition and call are in different files, compiler

needs more information to perform checks this is done through function prototypes


62

5/9/2012

Function Prototypes (cont’d)

Prototype looks like 1st line of function definition type

name

argument types

float dp(int n, float *x, float *y);

Argument names are optional:

float dp(int, float*, float*);


63

5/9/2012

Function Prototypes (3)

Prototypes are often contained in include files

/* mycode.h contains prototype for myfunc */

#include “mycode.h”

int main(){

…

myfunc(x);

…

}


64

5/9/2012

Basics of Code Management

Large codes usually consist of multiple files

Some programmers create a separate file for each

function Easier to edit

Can recompile one function at a time

Files can be compiled, but not linked, using –c option;

then object files can be linked later

gcc –c mycode.c

gcc –c myfunc.c

gcc –o mycode mycode.o myfunc.o


65

5/9/2012

Exercise 8

Put dot-product function and main program in separate

files

Create header file function prototype

.h suffix

include at top of file containing main

Compile, link, and run

solution


66

5/9/2012


Makefiles

make is a Unix utility to help manage codes

When you make changes to files, it will automatically deduce which files have been modified and compile

them

link latest object files

Makefile is a file that tells the make utility what to do

Default name of file is “makefile” or “Makefile” Can use other names if you’d like


67

5/9/2012

Makefiles (cont’d)

Makefile contains different sections with different

functions The sections are not executed in order!

Comment character is # As with source code, use comments freely


68

5/9/2012

Makefiles (3)

Simple sample makefile

### suffix rule

.SUFFIXES:

.SUFFIXES: .c .o

.c.o:

gcc -c $*.c

### compile and link

myexe: mymain.o fun1.o fun2.o fun3.o

gcc –o myexe mymain.o fun1.o fun2.o fun3.o


69

5/9/2012

Makefiles (4)

Have to define all file suffixes that may be

encountered

.SUFFIXES: .c .o

Just to be safe, delete any default suffixes first with a

null .SUFFIXES: command

.SUFFIXES:

.SUFFIXES: .c .o


70

5/9/2012

Makefiles (5)

Have to tell how to create one file suffix from another

with a suffix rule

.c.o:

gcc -c $*.c

The first line indicates that the rule tells how to create

a .o file from a .c file

The second line tells how to create the .o file

*$ is automatically the root of the file name

The big space before gcc is a tab, and you must use it!


71

5/9/2012

Makefiles (6)

Finally, everything falls together with the definition of a

rule

target: prerequisites

recipe

The target is any name you choose Often use name of executable

Prerequisites are files that are required by other files e.g., executable requires object files

Recipe tells what you want the makefile to do

May have multiple targets in a makefile


72

5/9/2012

Makefiles (7)

Revisit sample makefile

### suffix rule

.SUFFIXES:

.SUFFIXES: .c .o

.c.o:

gcc -c $*.c

### compile and link

myexe: mymain.o fun1.o fun2.o fun3.o

gcc –o myexe mymain.o fun1.o fun2.o fun3.o


73

5/9/2012

automatic variable for file root

Makefiles (8)

When you type “make,” it will look for a file called

“makefile” or “Makefile”

searches for the first target in the file

In our example (and the usual case) the object files

are prerequisites

checks suffix rule to see how to create an object file

In our case, it sees that .o files depend on .c files

checks time stamps on the associated .o and .c files to

see if the .c is newer

If the .c file is newer it performs the suffix rule In our case, compiles the routine


74

5/9/2012

Makefiles (9) Once all the prerequisites are updated as required, it

performs the recipe

In our case it links the object files and creates our

executable

Many makefiles have an additional target, “clean,” that

removes .o and other files clean:

rm –f *.o

When there are multiple targets, specify desired target

as argument to make command make clean


75

5/9/2012

Makefiles (10)

Also may want to set up dependencies for header files When header file is changed, files that include it will automatically

recompile

example: myfunction.o: myincludefile.h

if time stamp on .h file is newer than .o file and .o file is required in

another dependency, will recompile myfunction.c

no recipe is required


76

5/9/2012

Exercise 9a

Create a makefile for your dot product code

Include 2 targets create executable

clean

Include header dependency (see previous slide)

Delete old object files and executable manually rm *.o dotprod

Build your code using the makefile

solution


77

5/9/2012


Exercise 9b

Type make again should get message that it’s already up to date

Clean files by typing make clean Type ls to make sure files are gone

Type make again will rebuild code

Update time stamp on header file touch dp.h

Type make again should recompile main program, but not dot product function


78

5/9/2012

C Preprocessor

Initial processing phase before compilation

Directives start with #

We’ve seen one directive already, #include simply includes specified file in place of directive

Another common directive is #define #define NAME text

NAME is any name you want to use

text is the text that replaces NAME wherever it appears in source

code


79

5/9/2012

C Preprocessor (cont’d)

#define often used to define global constants

#define NX 51

#define NY 201

…

float x[NX][NY];

Also handy to specify precision

#define REAL double

…

REAL x, y;


80

5/9/2012

C Preprocessor (3)

Can also check values using the #if directive

In the current exercise code, the function fabsf is used,

but that is for floats. For doubles, the function is fabs.

We can add this to dp.h file:

#if REAL == double

#define ABS fabs

#else

#define ABS fabsf

#endif


82

5/9/2012

C Preprocessor (4)

scanf format “%f” for 4-byte floats

“%lf” (long float) for 8-byte floats

Can also use a directive for this:

#if REAL == double

#define SCANFORMAT “%lf”

#else

#define SCANFORMAT “%f”

#endif


83

5/9/2012

C Preprocessor (5)

#define can also be used to define a macro with

substitutable arguments

#define ind(m,n) (n + NY*m)

k = 5*ind(i,j); k = 5*(i + NY*j);

Be careful to use ( ) when required! without ( ) above example would come out wrong

k = 5*i + NY*j wrong!


84

5/9/2012

Exercise 10 Modify dot-product code to use preprocessor

directives to declare double-precision floats Add directives to header file to define REAL as shown in “C

Preprocessor (cont’d.)”

Add directives to header file to choose ABS as shown in “C

Preprocessor (3)”

Add directives to header file to choose SCANFORMAT as shown in

“C Preprocessor (4)”

Change all occurrences of float to REAL in dotprod.c, dp.c, and dp.h

Change fabsf to ABS in main routine

Change “%f” (including quotes) to SCANFORMAT in main

Include math.h in main program if you have not already done so

Include dp.h in function dp.c (for definition of REAL)

solution


85

5/9/2012


Structures

Can package a number of variables under one name

struct grid{

int nvals;

float x[100][100], y[100][100], jacobian[100][100];

};

Note semicolon at end of definition


86

5/9/2012

Structures (cont’d)

To declare a variable as a struct

struct grid mygrid1;

Components are accessed using . mygrid1.nvals = 20;

mygrid1.x[0][0] = 0.0;

Handy way to transfer lots of data to a function

int calc_jacobian(struct grid mygrid1){…


87

5/9/2012

Exercise 11

Define struct rvec with 2 components in your header

file (.h) vector length (int)

pointer to REAL vector

Modify dot-product code to use rvec structure

solution


88

5/9/2012


i/o

Often need to read/write data from/to files rather than

screen

File is associated with a file pointer through a call to

the fopen function

File pointer is of type FILE, which is defined in stdio.h.


89

5/9/2012

i/o (cont’d)

fopen takes 2 character-string arguments file name

mode

“r” read

“w” write

“a” append

FILE *fp;

fp = fopen(“myfile.d”, “w”);


90

5/9/2012

i/o (3)

Write to file using fprintf Need stdio.h

fprintf has 3 arguments 1. File pointer

2. Character string containing what to print, including any formats

%f for float or double

%d for int

%s for character string

3. Variable list corresponding to formats


91

5/9/2012

i/o (4)

Special character \n produces new line (carriage

return & line feed) Often used in character strings

“This is my character string.\n”

Example:

fprintf(fp, “x = %f\n”, x);

Read from file using fscanf arguments same as fprintf

When finished accessing file, close it

fclose(fp);


92

5/9/2012

Exercise 12

Modify dot-product code to write the dot-product result

to a file

If magnitude is small, still write message to screen

rather than file

After result is written to file, write message “Output

written to file” to screen.

solution


93

5/9/2012


Binary i/o Binary data require much less disk space than ascii

(formatted) data

Use “b” suffix on mode

fp = fopen(“myfile.d”, “wb”);

Use fwrite, fread functions

float x[100];

fwrite( x, sizeof(float), 100, fp )

Note that there is no format specification

We’re strictly writing data


94

5/9/2012

pointer to

1st element no. bytes in

each element max. no. of

elements

file pointer

Exercise 13

Modify dot-product program to: Write result to binary file

just write value, not character string

After file is closed, open it back up and read and print result to make

sure that it wrote/read correctly

don’t forget to open file with “rb” rather than “wb”

solution


95

5/9/2012


Command-Line Arguments It’s often convenient to type some inputs on the

command line along with the executable name, e.g.,

mycode 41.3 “myfile.d”

Define main with two arguments: int main(int argc, char *argv[ ])

1. argc is the number of items on the command line, including name

of executable

• “argument count”

2. argv is an array of character strings containing the arguments

“argument values”

argc[0] is pointer to executable name

argc[1] is pointer to 1st argument, argc[2] is pointer to 2nd

argument, etc.


96

5/9/2012

Command-Line Arguments (cont’d)

Arguments are character strings, often want to convert

them to numbers

Some handy functions: atoi converts string to integer

atof converts string to double

To convert to float, recast result of atof

They live in stdlib.h

arguments are pointers to strings, so you would use, for example

ival = atoi(argv[2])

to convert the 2nd argument to an integer


97

5/9/2012

Command-Line Arguments (3)

Often want to check the value of argc to make sure the

correct number of command-line arguments were

provided

If wrong number of arguments, can stop execution

with exit statement Can exit with status, e.g.:

exit(1);

With bash shell, view status by echoing ‘$?’:

$ echo $?

1


98

5/9/2012

Exercise 14 Modify dot-product code to enter the vector length as a

command-line argument rather than prompting for it

Use atoi

Add test on argc to make sure a command-line

argument was provided argc should equal 2, since the executable name counts

if argc is not equal to 2, print message and return to stop execution

solution


99

5/9/2012


References

Lots of books available Kernighan & Ritchie, “The C Programming Language”

gcc http://gcc.gnu.org/onlinedocs/gcc-4.5.1/gcc/

If you’d like to move on to C++ Good C++ book for scientists:

Barton and Nackman, “Scientific and Engineering C++”

Quick and dirty C++ book:

Liberty, “Teach Yourself C++ in 21 Days”


100

5/9/2012

http://gcc.gnu.org/onlinedocs/gcc-4.5.1/gcc/



Survey

Please fill out the course survey at

http://scv.bu.edu/survey/tutorial_evaluation.html


101

5/9/2012

integrating r and c/c++ - boston · pdf filercpp, rcpparmadillo, rcppbugs, etc. applications...

Documents